Developing in vivo assays for investigation of p75NTR and NRH1 transmembrane domain cleavage events using zebrafish embryos

γ-secretase is an important protease complex responsible for the cleavage of over 100 substrates within their transmembrane domains. γ-secretase acts in Alzheimer’s disease by cleavage of AMYLOID BETA (A4) PRECURSOR PROTEIN to produce aggregation-prone Amyloid beta peptide. Other γ-secretase substrates such as p75NTR are also relevant to Alzheimer’s disease. How γ-secretase cleavage site specificity is determined is still unclear. A previous study using Xenopus laevis to investigate the proteolytic processing of p75NTR and its homolog NRH1 found that transmembrane cleavage of NRH1 was not sensitive to the γ-secretase inhibitor DAPT, suggesting that it is not processed by γ-secretase. To investigate this further, we identified zebrafish orthologues of the genes p75NTR and NRH1 and developed in vivo assays to assess cleavage of the resultant p75NTR and Nrh1 proteins. Our observations from these assays in zebrafish are consistent with the Xenopus laevis study. Inhibition of γ-secretase by DAPT treatment results in accumulation of uncleaved p75NTR substrate, while cleavage of Nrh1 is not affected. This supports that p75NTR is cleaved by γ-secretase while Nrh1 is cleaved by a separate γ-secretase-like activity. We extended our approach by generating a chimeric Nrh1 protein in which the Nrh1 transmembrane domain was replaced by that of p75NTR, in an attempt to determine whether it is the p75NTR TMD that confers susceptibility for γ-secretase cleavage. Our results from analysis of this chimeric protein revealed that the p75NTR transmembrane domain alone is insufficient to confer γ-secretase cleavage susceptibility. This is not completely unexpected, as there is evidence to suggest that other factors are crucial for selection/cleavage by the γ-secretase complex. We have established a system in which we can now attempt to dissect the structural basis for γ-secretase cleavage specificity and evolution.

Introduction 36 γ-secretase is a multi-subunit membrane-bound aspartyl protease complex responsible for 37 cleavage of over 100 substrates including Amyloid Precursor Protein (APP), Notch and the 38 p75 neurotrophin receptor (p75 NTR ) [1]. γ-secretase has been identified as a member of the 39 intramembrane cleaving protease family (I-CLiP). I-CLiPs cleave type 1 membrane proteins 40 enzymatically via a process termed regulated intramembrane proteolysis (RIP) [2,3]. The 41 most studied function of γ-secretase is processing of APP. This is due to the seemingly 42 critical role of APP in Alzheimer's disease (AD) etiology. Although numerous publications 43 have discussed the perceived role of γ-secretase in AD, the specific nature of substrate 44 selection by this protease is not yet clearly defined. 45 γ-secretase substrates are typically derived from large precursor proteins that undergo a 46 prerequisite removal of their ectodomain/lumenal domain prior to γ-secretase cleavage [3]. 47 Early research suggested that the only other prerequisite was that the substrate must be a type 48 1 transmembrane protein [4]. Later studies showed that additional factors may guide substrate 49 selection by γ-secretase. It has previously been suggested that dimerisation of substrates 50 and/or the structure of substrate α-helices may regulate γ-secretase activity [5]. Also, γ-51 secretase substrate recognition and cleavage is much more efficient for ectodomains with 52 fewer than ~50 remaining amino acid residues [6]. Previous studies have attempted to define 53 the amino acid residues of Notch and APP required for γ-secretase cleavage [7]. However, a 54 distinct cleavage recognition site for γ-secretase within the transmembrane domains (TMDs) 55 of its target proteins has not been defined [2]. 56 p75 NTR , also known as the 'low-affinity nerve growth factor receptor' (LNGFR), is one of the 57 many substrates of γ-secretase subject to cleavage within its transmembrane domain [8]. 58 p75 NTR has been implicated in neuronal survival, myelination and neurite outgrowth among 4 59 other pathways during vertebrate nervous system development, through its interactions with 60 neurotrophins and Trk receptors [9]. Also, the Aβ peptide can act as a ligand for p75 NTR and 61 is proposed to play a role in cholinergic neuron loss, implicating this protein in AD [10,11]. 62 Kanning et al (2003) investigated the proteolytic processing of p75 NTR , along with what they 63 described as the "neurotrophin receptor homologs" (NRH), NRH1 and NRH2 [8]. Database 64 and EST searches have established that the genes coding for these two proteins show greater 65 sequence similarity to p75 NTR than to any other homologous genes [8]. Experiments by 66 Kanning et al (2003) confirmed processing of p75 NTR by α-secretase (ADAM10) and γ-67 secretase. Western blot analysis of NRH1 and NRH2 indicated that both proteins are cleaved 68 within their transmembrane domains [8]. However, Kanning et al (2003) observed that the 69 commonly used inhibitor of γ-secretase activity, DAPT, had no effect on the cleavage of 70 NRH1 or NRH2, suggesting that transmembrane cleavage of these proteins is not by γ-71 secretase [8]. 72 The possible lack of sensitivity to γ-secretase inhibitors of the p75 NTR homologs NRH1 and 73 NRH2 may have interesting applications. p75 NTR and its homologs share high sequence 74 similarity in their transmembrane domain, where γ-secretase cleavage occurs [8] and 75 comparison of the transmembrane domains of p75 NTR and NRH1 might allow definition of 76 transmembrane domain characteristics critical to permit cleavage by γ-secretase. We have 77 developed an in vivo zebrafish assay that can be used to investigate the structural differences 78 in the transmembrane domains of these two proteins that cause their differential sensitivity to 79 γ-secretase. Our results support the observations made by Kanning et al (2003)  considered to be significant. We did not remove any outliers from the analysis.  can be found in Table S1. Branchiostoma floridae (lancelet) was used as an out-group as this 211 was the most distant relative to zebrafish that returned a result when conducting tblastn 212 searches using human p75 NTR and X. laevis NRH1. Interestingly, tblastn searches of the 213 lancelet genome using both p75 NTR and NRH1 returned the same gene in lancelet (Table S1).

214
As the chicken genome contains both p75 NTR and NRH1-like sequences, tblastn searches of 215 the lancelet genome using both full-length chicken sequences were performed to confirm the 216 preliminary findings. These searches returned results identical to those using human p75  It was difficult to discern which of the remaining three possible zebrafish p75 NTR protein 230 sequences was most likely to represent the true p75 NTR in zebrafish from the phylogenetic 231 analyses, although the sequence that we designated zebrafish p75-like-chr3 appeared to be 232 marginally more similar to human p75 NTR (Fig 1). The viral 2A (v2A) peptide ribosomal-skip mechanism allows for expression of two different 295 proteins independently of one another. The skip mechanism occurs within the v2A sequence 296 when a peptide bond fails to form between the penultimate (glycine) and final (proline) 297 residue. Translation continues despite this failure, and tandem protein products are produced 298 in a stoichiometric manner [21]. In order to express truncated p75 NTR or Nrh1 simultaneously 299 with an internal reference standard from the same expression vector, we included a v2a 300 sequence at the C-terminal of ssFLAG-p75 NTR C201-dGFP and ssFLAG-Nrh1C201-dGFP 301 followed by coding sequence for the red fluorescence protein, mCherry. We describe these 302 Tol2-based expression constructs as pT2ALssFLAG-p75 NTF C201-dGFP-v2a-mCherry and 303 pT2ALssFLAG-Nrh1C191-dGFP-v2a-mCherry (Fig 2, 1). For simplicity, we will henceforth 304 refer to them as p75 NTF C201-dGFP and Nrh1C191-dGFP respectively. This design enables 305 stoichiometric production of p75 NTF C201-dGFP or Nrh1C191-dGFP simultaneously with 306 mCherry, allowing for normalisation of protein expression between successive batches of 307 injected embryos. observed for p75 NTF C201-dGFP-injected embryos treated with DAPT (Fig 3, A), suggesting 342 that zebrafish p75 NTR is, indeed, processed by γ-secretase. However, when western 16 343 immunoblot densitometry data was normalised to mCherry across three replicates, the p-344 value of this observed increase (p = 0.1697) did not support the likelihood that uncleaved 345 zebrafish p75 NTR is consistently accumulated when treated with DAPT (Fig 3, B). This 346 accumulation of substrate was not observed when Nrh1C191-dGFP was subjected to DAPT 347 treatment (Fig 3, A), suggesting that zebrafish Nrh1 is not sensitive to the γ-secretase 348 inhibitor DAPT. This result supports the observation made by Kanning et al (2003)   analysis (after confirmation that they were expressing GFP observable by its fluorescence). An apparent trend of accumulation of p75 NTF C201-dGFP due to γ-secretase inhibition was 384 observed by western immunoblotting. However, statistical analysis of the densitometry 385 measurements did not indicate significance due to the considerable variability between 386 samples (Fig 3, B and C). A contributor to this variability may have been the necessity to 387 strip and re-probe the western blot with the anti-mCherry antibody. To overcome this, the red 388 fluorescence gene mCherry was excised from the constructs and replaced with a second GFP 389 gene downstream of the C-terminal of v2a, producing vectors pT2ALssFLAG-p75 NTF C201-390 dGFP-v2a-GFP and pT2ALssFLAG-Nrh1C191-dGFP-v2a-GFP (Fig 2, 2). For simplicity, 18 391 we will henceforth refer to these as p75 NTF C201-dGFPx2 and Nrh1C191-dGFPx2 392 respectively. This minor adjustment in construct design allows for the internal expression 393 standard to be visualised using the same anti-GFP antibody as detects the p75 NTF C201-dGFP 394 and Nrh1C191-dGFP fusions.

395
To evaluate the effectiveness of the modified assay constructs we performed injections on 396 numerous batches of embryos and then ran protein samples on multiple western blots.

397
Analysis using the new assay constructs consistently displayed an increase in accumulation of 398 the p75 NTF C201-dGFPx2 substrate when treated with the γ-secretase inhibitor DAPT. This 399 result was confirmed statistically by combining band intensity data from across all western 400 blots and performing a two-tailed t-test assuming unequal variances, resulting in a p value of 401 0.0047 (Fig 4, A). Conversely, there was no observed increase in Nrh1C191-dGFP substrate accumulation in 410 response to DAPT treatment. Statistical analysis found no significant difference between the 411 treated and untreated samples (p =0.9037) (Fig 4, B). Although the p75 NTR western 412 immunoblot data was similar across numerous blots there was a high degree in variability in 413 the normalised values across Nrh1 immunoblots. This variability was unexpected and seems 414 to be due to variation in the amount of free GFP on each blot. It is possible that GFP stability The sequence similarity of p75 NTR and its homolog Nrh1 imply that these two genes share a

429
To begin dissection of γ-secretase cleavage substrate specificity using our assay, we designed 430 a chimaeric construct in which Nrh1's transmembrane domain was replaced with the 431 transmembrane domain from p75 NTR . The new construct, termed pT2ALssFLAG-A2C-432 dGFP-v2a-GFP (Fig 2, 3), simplified to A2C-dGFPx2, was injected into one cell stage 433 embryos which were subsequently treated as previously with or without DAPT. Protein 434 samples were then collected at 24 hpf for analysis by western immunoblot. This did not 435 reveal an accumulation of substrate when γ-secretase was inhibited (Fig 5) and multi pass membrane proteins in some cases [38,39]. If we assume that NRH1 is a type 1 517 transmembrane protein like its homologue p75 NTR , then we can reasonably exclude two of the 518 above classes of membrane cleaving proteases as candidates, namely, SPP (and SPP-like) and 519 S2P (including S2P-like). However, the orientation of NRH1 within the membrane has not 520 yet been investigated. Our assay could be used to test a range of protease inhibitors to 521 identify which enzyme(s) cleave Nrh1.

522
The question of the effect of the α-helix structure of p75 NTR or NRH1 TMD on their cleavage 523 susceptibility has not yet been investigated. Previous observations from a study of the effects 524 of the TMD of Gurken found that a β-sheet conformation was required for cleavage by the 525 archaeal PRESENILIN homologue, MCMJR1. This raised the question of whether NRH1 526 can interact with PRESENILIN, but perhaps due to its TMD being in an α-helical 527 conformation, cannot be cleaved by it. A simple amino acid sequence alignment of the 528 zebrafish p75 NTR and Nrh1 TMDs (Fig 6) revealed that, while the p75 NTR TMD carries a 529 proline residue that would supposedly allow it flexibility to conformationally change between 530 an α-helix and β-sheet, the Nrh1 TMD lacks this residue. Interestingly, it has previously been 531 observed that insertion of a single proline into a TMD can trigger cleavage of normally un-532 cleavable TMD's [30]. It may, therefore, be of future interest either to insert or substitute a 533 proline residue into the Nrh1 TMD to investigate the effect on its γ-secretase cleavage 534 susceptibility. The results of such an experiment using our established assay may provide an 535 answer for why Nrh1 is not naturally a γ-secretase substrate, while also contributing to the 536 understanding of γ-secretase cleavage susceptibility. Regarding the previously observed 24 537 cleavage of Xenopus NRH1 within its TMD and the question of what protease might be 538 responsible for this cleavage, it has been observed that all intramembrane cleaving proteases 539 (iCLIPs) prefer to cleave TMDs in their β-strand conformation [40]. If NRH1 is unable to 540 enter this conformation, perhaps there is some other unknown enzyme responsible for this 541 cleavage. If we wish to understand the cleavage properties of NRH1, it will be important to 542 further investigate the conformational state of its TMD.   Table S2. Intensity ratios from western immunoblots for Figure 3 708 Table S3. Intensity ratios from western immunoblots for Figure 4 709 Table S4. Intensity ratios from western immunoblots for Figure 5 710 Text File S1. Construct overviews and full sequences